Ground cover classification is one of the core applications in remote sensing. Classification of imaging radar data using conventional single-channel or dual-channel usually results in poor accuracy due to limited number of observations. For this reason, applications of full polarimetric data are growing. In this article, Radarsat-2 quad-pol data of an agriculture scene acquired on August, 2013 over the Nile Delta, Egypt, has been used to evaluate the accuracy of three supervised classification methods, Support Vector Machine (SVM), multilayer perceptron (MLP), and decision tree (DT), in discriminating between four different crops: rice, maize, grape and cotton, in addition to urban land cover. A set of principal components was used in the SVM classifier. Our goal is to build a generic model that minimizes the error of future classification using the same training data. Results indicate that combining the most commonly used polarimetric parameters from Pauli, Cloude–Pottier, and Freeman–Durden decompositions along with the three fundamental backscatter coefficients ( , , and ) using SVM classifier strike an appropriate balance that minimizes the effect of the training error on the classification accuracy without going to over-fitting classification. This leads to the highest overall classification accuracy of 94.48%. The overall classification accuracy using different combinations of parameters and different classifiers is given.
The sustainable management of water resources and the development of climate-resilient infrastructure depend on the precise identification of water bodies in satellite imagery. This paper presents a novel deep learning architecture that integrates a convolutional block attention module (CBAM) into a modified EfficientNet–UNet backbone. This integration allows the model to prioritize informative features and spatial areas. The model robustness is ensured through a rigorous training regimen featuring five-fold cross-validation, dynamic test-time augmentation, and optimization with the Lovász loss function. The final model achieved the following values on the independent test set: precision = 90.67%, sensitivity = 86.96%, specificity = 96.18%, accuracy = 93.42%, Dice score = 88.78%, and IoU = 79.82%. These results demonstrate improvement over conventional segmentation pipelines, highlighting the effectiveness of attention mechanisms in extracting complex water-body patterns and boundaries. The key contributions of this paper include the following: (i) adaptation of CBAM within a UNet-style architecture tailored for remote sensing water-body extraction; (ii) a rigorous ablation study detailing the incremental impact of decoder complexity, attention integration, and loss function choice; and (iii) validation of a high-fidelity, computationally efficient model ready for deployment in large-scale water-resource and ecosystem-monitoring systems. Our findings show that attention-guided segmentation networks provide a robust pathway toward high-fidelity and sustainable water-body mapping.